GAFD-CC: Global-Aware Feature Decoupling with Confidence Calibration for OOD Detection

TL;DR

GAFD-CC introduces a global-aware feature decoupling and confidence calibration method to improve out-of-distribution detection by refining decision boundaries and leveraging feature-logit correlations.

Contribution

It proposes a novel global-aware feature decoupling technique guided by classification weights combined with confidence calibration for enhanced OOD detection.

Findings

Achieves state-of-the-art performance on large-scale benchmarks.

Effectively refines decision boundaries for better ID/OOD separation.

Demonstrates strong generalization across diverse datasets.

Abstract

Out-of-distribution (OOD) detection is paramount to ensuring the reliability and robustness of learning models in real-world applications. Existing post-hoc OOD detection methods detect OOD samples by leveraging their features and logits information without retraining. However, they often overlook the inherent correlation between features and logits, which is crucial for effective OOD detection. To address this limitation, we propose Global-Aware Feature Decoupling with Confidence Calibration (GAFD-CC). GAFD-CC aims to refine decision boundaries and increase discriminative performance. Firstly, it performs global-aware feature decoupling guided by classification weights. This involves aligning features with the direction of global classification weights to decouple them. From this, GAFD-CC extracts two types of critical information: positively correlated features that promote in-distribution (ID)/OOD boundary refinement and negatively correlated features that suppress false positives and tighten these boundaries. Secondly, it adaptively fuses these decoupled features with multi-scale logit-based confidence for comprehensive and robust OOD detection. Extensive experiments on large-scale benchmarks demonstrate GAFD-CC's competitive performance and strong generalization ability compared to those of state-of-the-art methods.